Funding is computed from the gap between the perpetual and its index, and venues clamp the result within a maximum. The cap exists so a violent dislocation cannot produce a payment large enough to wipe out accounts through funding alone.
Sensible protection, and it creates a blind spot.
What happens when the cap binds
Funding stops varying. The perpetual can trade further and further from the index and the printed rate does not move, because it is already at its ceiling.
So at exactly the moment positioning is most extreme, the number you would use to measure that extremity goes flat. Reading it as “crowding stopped increasing” is precisely backwards — it stopped being measurable.
How to notice
Compare the funding rate against the venue’s stated cap. If it is pinned at the maximum across consecutive intervals, treat the figure as a floor on crowding rather than a measurement of it.
And check the basis directly — the perpetual against the index — since that keeps moving after funding has stopped. It is the uncapped version of the same signal, and the relationship is the one in spot-perp basis.
Why it matters for the percentile
A capped series wrecks any ranking computed on it. A distribution with a hard ceiling will show many observations sitting exactly at the maximum, so a percentile cannot distinguish between them — every extreme episode looks identical.
Which is a specific reason the funding extremes discipline needs care: the tail of the distribution is compressed by construction, not by the market.
The general lesson
Know the bounds of any figure before ranking it. A clamped series looks like data and behaves like data right up to the clamp, where it silently stops carrying information — and that is the region you were most interested in.